The behavior of the readResultSetHeaderPacket differs from that of the MySQL client

[ShenFeng312]

// http://dev.mysql.com/doc/internals/en/com-query-response.html#packet-ProtocolText::Resultset
func (mc *okHandler) readResultSetHeaderPacket() (int, error) {
	// handleOkPacket replaces both values; other cases leave the values unchanged.
	mc.result.affectedRows = append(mc.result.affectedRows, 0)
	mc.result.insertIds = append(mc.result.insertIds, 0)

	data, err := mc.conn().readPacket()
	if err == nil {
		switch data[0] {

		case iOK:
			return 0, mc.handleOkPacket(data)

		case iERR:
			return 0, mc.conn().handleErrorPacket(data)

		case iLocalInFile:
			return 0, mc.handleInFileRequest(string(data[1:]))
		}

		// column count
		num, _, n := readLengthEncodedInteger(data)
		if n-len(data) == 0 {
			return int(num), nil
		}

		return 0, ErrMalformPkt
	}
	return 0, err
}

In Go projects, we check the packet's length. However, in the MySQL client and other language drivers, this is not done. They only extract the necessary content from the packet without checking its length. This approach provides the MySQL protocol with sufficient flexibility. Therefore, I believe we should remove the length check here.

		if n-len(data) == 0 {
			return int(num), nil
		}

[ShenFeng312]

I attempted to add some additional content to this packet, and it threw the following exception:

However, I can obtain the correct results on other clients and drivers.